SLOP CANNONS

The meeting had been underway for nine minutes before Marc Aldrin understood that nothing useful was going to happen.

The Director of Security stood at the front of the conference room, working through prepared remarks with the practiced cadence of someone who had delivered versions of them before. Seven minutes in, she began publicly attributing the work of a senior engineer to herself. Not subtly. Not through omission. She named the project, described the technical decisions, and used the word “I” six times in ninety seconds to describe a system she had not ideated, designed, built, nor could operate. In fact, she had actively blocked resources for the work during the two quarters it took the engineer to ship it anyway. She did this while the engineer was in the room. Same table, four seats from Marc. Everyone saw it. No one said a word. She was close with a chief, and closeness with a chief was a currency that outranked competence, integrity, and the forty-page values document that HR had distributed at the last all-hands. A document that espoused the culture and lived in a Slack channel with nine hundred and eighty members that hadn’t seen a message since last October. The head of HR had deemed any human who was not the head of HR unfit to post in the human resources Slack channel. Nobody found this ironic. Or if they did, they had stopped saying so.

Marc sat three chairs from the head of the table with a cooling mug of coffee and the familiar sensation that an expensive amount of talent was being converted, minute by minute, into theater. On the wall behind the Director of Security, a tiled Zoom grid of remote attendees watched in silence. Most of Marc’s team was up there. So were the engineers from adjacent groups, the product leads, the analysts. Each muted rectangle lending the call the flattened, ceremonial quality of an event that existed more to be witnessed than to decide anything.

The company made enterprise software. Respectable software. Expensive software. The kind bought by large organizations that needed their sales pipelines, customer data, and revenue operations arranged into something legible enough to present to boards and auditors. It was not an unserious business. Which was why Marc still could not quite believe that, four weeks earlier, the executive team had decided the future of this otherwise ordinary CRM company involved launching a cryptocurrency token.

A rewards token, they called it.

Close a deal, earn tokens. Use tokens to unlock premium analytics, partner incentives, or whatever else product marketing could turn into a slide. The CEO had watched a sitting president of the United States shill a memecoin from the Oval Office and decided that was permission. The proposal moved from idea to mandate in six weeks. Compliance received a deck titled “regulatory alignment in progress.” Engineering received a date. Legal, as far as Marc could tell, received nothing at all. A CRM company was minting coins while its own engineers filed expense reports through a system that crashed every other Friday.

The project manager took his turn. He opened with a string of acronyms that landed on the room like light rain on concrete. OKRs. KPIs. SLAs. A priority-zero incident that had been priority-zero for six weeks and would remain priority-zero for six more, because naming a thing is not the same as fixing it. Eyes glossed before his first run-on sentence ended.

Marc looked around the room. Half the C-suite was present. Not the half that made hard decisions. The other half. The half that liked visibility. The half that preferred to be near action without ever having to own the consequences of it. Between the executives in the room, the executives on video, and the managers beneath them, Marc guessed there were several million dollars of annual salary listening to a meeting that could have been condensed into three sentences and a deadline.

His mug was from a conference in Chicago nearly a decade ago, back when he still wrote code every day. He kept it on his desk for reasons he never explained well, though privately he thought of it the way other people thought of an old photograph. Evidence from a former life. Proof that he had once spent more time building things than talking about the people building them.

Now he ran five engineering managers. Under them sat teams responsible for pipeline automation, contact graph ingestion, integrations with every payments provider and marketing platform their enterprise clients demanded, and the thankless machinery of data residency compliance across forty countries with forty different privacy regimes. Their systems processed billions of messages a day. They had not suffered a meaningful outage in two years. Most weeks, the company ran on the back of infrastructure whose existence senior leadership remembered only when they wanted something impossible by Friday.

His boss, the Director of Engineering, began to speak.

Marc listened with the weary attentiveness of a man who had heard many forecasts of clear skies while standing in floodwater. The DoE said the right things in the right order. The team needed to improve. Communication had to tighten. Ownership needed to be clearer. He did not say at what. He never did. Specificity would have required him to know what the teams were building, and knowing what the teams were building would have required him to attend a meeting that wasn’t this one. He had a chief of staff who converted his abstractions into Jira epics and a calendar blocked with syncs he attended on mute. He requested headcount and asked the occasional question, always in front of the CTO, that proved he had read the executive summary of a document his chief of staff had summarized from a document an engineer had written.

In many organizations, directors at this level operate as a firewall. Marc’s Director of Engineering was no different. Inbound traffic from leadership passed through unfiltered—every demand, every deadline, every mandate arrived at the engineering layer at full force. Outbound traffic was another matter. Deep packet inspection on every message headed upstream. Anything containing structural failure, inconvenient truth, or a request that would require real introspection from the layers above was silently dropped. The executive layer saw only clean, sanitized telemetry. Green dashboards. He existed at that boundary because the network architecture required a security appliance between the people who funded the work and the people who understood it.

There would be a retrospective next Wednesday.

Marc had nothing against retrospectives in principle. In healthy systems, they were useful. In unhealthy ones, they were a filing cabinet for truth no one intended to act on. Leadership had read a snippet on X about better outcomes from documenting failures. Nobody mentioned that no one had read the last four postmortems. Nobody mentioned that the systemic issues identified in those postmortems had never received a single resource for remediation. There was no time. There was a backlog of thirty-six months of feature work that the roadmap committee had approved, the quarterly planning process had blessed, and that the organization would spend the next six weeks re-planning before declaring the plan obsolete by week two of execution.

He knew the pattern. Action items would be written on a whiteboard or tucked into a document. Owners would be assigned. Deadlines would be discussed with optimism no one in the room believed. By the following week, the same people would still be carrying the same impossible loads, and the newly documented fixes would settle quietly into the growing sediment of institutional good intentions.

The retrospective was not a solution. It was exhaust from a machine built so heavy with its tiers and so segmented by its self-imposed verticals that the weight of itself prevented any real value creation. But the failures would be documented for the next grand rearchitecture of the org chart coming in twenty-four months.

The organization had become exceptionally skilled at describing its problems. Solving them was another matter.

That was what had begun to trouble Marc most. Not the chaos. Not even the waste. He had worked in large systems long enough to expect both. It was the growing realization that competence inside this kind of environment often meant becoming fluent in the maintenance of unreality. You learned to translate frustration into language safe for leadership review. You learned to describe chronic failure as temporary complexity. You learned to assign owners to structural problems no individual could solve. If you did it well enough, people called you effective.

Marc was effective. He was good at every version of the job he had ever held. Good at building teams. Good at restoring order. Good at turning ambiguity into execution. Good at absorbing pressure from above and redistributing it into plans, headcount requests, and careful language.

Lately, though, a thought had begun to return to him in quiet moments, unwelcome and increasingly difficult to dismiss.

What if being good at this version of the job was the problem?

The stories in this book are fiction. They are drawn from real experience, mine, good friends, and the incredible people I have worked alongside. It is exaggerated to prove a point, the way a golf instructor tells you to overcorrect until the movement feels absurd, because the real swing lives between the old habit and the overcorrection. The scenes are fiction. The patterns are documentary.

This is not a book about doom. It is not a book about AI replacing humans. It is not a prediction that engineering leadership is going away. Engineering leadership matters more now than it ever has. What is going away is the ceremony around it. The layers that masquerade as leadership. The title inflation that substitutes for impact. Strip those away and what remains is the real work: lifting an entire organization. Making a junior engineer dangerous. Making a non-technical expert a builder. Unlocking capability in people who were never given the tools or the trust to discover what they could do. AI enables this. Not fewer leaders. Better ones.

This is not an indictment of individuals.

Individuals matter, of course. They set tone, they make choices, they protect or damage the people around them. But most of the people inside these systems are not acting as aberrations. They are adapting, with varying degrees of grace, to structures that reward legibility upward more reliably than value outward. It is easy to make villains out of whoever happens to be speaking in the meeting. The project manager with his acronyms. The director who confuses composure with leadership. The executive who appears only when visibility is high and consequence is diffuse. They can all be frustrating. Some are destructive. But they are rarely the root cause. Remove one and another appears, shaped by the same pressures, speaking the same language, preserving the same machinery.

The machinery is the point. Modern engineering organizations are designed to make themselves intelligible to power before they make themselves useful to users. Headcount becomes a signal of seriousness. Complexity becomes a proxy for sophistication. Layers breed layers because each new layer can explain, coordinate, summarize, and defend the existence of the others. Before long, a meaningful share of institutional energy is spent not on building systems but on narrating them.

Sometimes brilliantly. That is part of the trap. These organizations can become extraordinarily articulate about their own intentions. They produce roadmaps, operating plans, strategy decks, postmortems, planning rituals, leveling matrices, architecture councils, executive reviews. They can describe the river in perfect detail while drifting farther from the water. And because the drift is gradual, it escapes notice until it has become culture.

A meeting ends without a decision but with a clean summary and a follow-up owner, so it feels productive. A postmortem finds the real issue, then assigns procedural remedies because structural remedies would require someone important to admit the problem is not local. A strong engineer spends more time preparing updates than improving the system. A manager is rewarded for reducing friction in the story rather than friction in the work. Architecture review boards staffed by people who haven’t touched production code in years, making decisions about systems they couldn’t operate. Calibration meetings where careers are negotiated by people who have never read the engineer’s code. Platform initiatives with eighteen-month timelines and success criteria so vague that failure is impossible because nobody defined success.

None of this is catastrophic in isolation. All of it compounds. Fiefdoms built on proximity, on shared hobbies, kids at the same school, houses in the same neighborhood, surviving reorg after reorg because closeness outranks competence. Castles in the sky, justified by slides nobody challenges because challenging the slide means challenging the VP who commissioned it. People learn to survive inside it. They learn which truths can be spoken plainly and which need translation into safer language. They learn to convert frustration into process. They learn that if a problem is old enough, it can be treated as weather. They learn that the official story and the private one are related but not identical.

If they stay long enough, many become fluent in both. That fluency is usually called maturity.

Sometimes it is. Sometimes it is simply adaptation to unreality.

I have been good at this game. I have built ladders. I have run calibrations. I have defended headcount I suspected was overhead because the system rewarded the defense and punished the question. I have succeeded inside these structures and been genuinely proud of some of what they made possible. Large organizations are not jokes. Many do important work.

But I have also watched capable builders leave. Not because they feared difficulty. Because they were tired of ornamental complexity pretending to be necessity.

What interests me most are the people who have not made peace with that adaptation. Some are young. Some are not. Some are senior enough to know exactly how the machine works and junior enough, in spirit, to still find it absurd. They are not united by ideology. They are united by proximity to the work. They can still feel the drag. They can still see how much effort is being burned to preserve appearances that no user, no customer, no production system has asked for.

They tend to ask impolite questions. What valuable thing became possible because this team exists? Why does this decision require this many witnesses? Why is so much interpretation needed between the person doing the work and the person funding it?

These are dangerous questions in institutions that survive through managed abstraction. Not because the questions are unfair. Because they are clarifying.

Everyone sees this. Anyone under thirty recognized the absurdity on day one. The director with twenty years sees it too. So does the VP who chairs calibrations. Most are far enough along to accept the dysfunction as cost of doing business. Mortgages. Vesting schedules. They learned to call compliance what is actually complacency.

The next generation will not. They see it, play it as long as it pays, and leave the moment it doesn’t. They are loyal to the work, not the org chart. They are leaving. Quietly. Steadily. And the organizations they leave behind are discovering that the middle was not load-bearing after all.

The old cover is thinning. That is why this book is direct. Not because outrage is the point. Outrage is cheap. Clarity is harder, and more useful.

This is changing. Generative AI is a forcing function, but it is not the only one. The refusal to pretend is the other.

Here is how I came to see it.

I did not arrive in software through the front gate. Before engineering, I spent years in sales. I studied finance in college. Dropped out. Watched older friends graduate with the same degree, land the jobs they were supposed to want, and loathe every hour of them while earning less than I was making being a proficient salesman. No reason to pay tuition to earn less.

So I sold. Inbound, outbound, the kind of selling that is ruthless about one thing: eventually, somebody has to receive something of value or the entire performance is exposed. You can hide behind vocabulary for a while. You can hide behind personality for a while. You cannot hide forever from the question of whether anything useful changed for the person on the other side of the table.

That question stayed with me. In 2010, I enrolled in community college and started studying computer science while working during the day. The startup boom was still gathering. Most people thought I was making a lateral move. I pivoted because software was the most powerful instrument for creating real change I had ever encountered. A user has a problem. You understand it, build the right thing, put it in front of them, and the world is concretely different by the end of the week. There is very little in modern work more satisfying than that. There is also very little more clarifying. Once you have felt that kind of leverage, it becomes difficult to respect measures of progress that float too far above it.

Two semesters in, a friend and I pitched TechStars on a semantic layer for websites. Metadata that let information find you instead of you searching for it. Push instead of pull. They loved the idea. Too early. Too costly. In hindsight, they were right. But the instinct was locked in: find the problem, build the thing, put it in front of someone, watch what happens.

That instinct shaped the environments I chose. I was drawn to inflection points. New systems. Hard transitions. Teams asked to build something meaningful before the institution had organized itself around the fact that it needed that thing. I chose organizations for freedom, not stability. The teams I built worked on things that mattered and thrived on the pressure of building something that communicated its own value. Not through a promotion packet. Through what it did and how it worked.

Over time I built from zero, inherited complexity, led small groups, led larger ones, and learned the difference between a structure that supports execution and one that feeds on it. One uses process to sharpen intent. Another uses process to distribute accountability until nobody can locate it. One makes hierarchy a tool. Another makes it the product.

That left me with a filter that has stayed with me longer than any management framework: does this increase the organization’s ability to create value, or has it become overhead defending its own right to exist?

Most organizations prefer proxies. Headcount growth. Story points. Velocity. PRs merged. Roadmap completion. None of these are meaningless. Some are occasionally useful. All of them become dangerous when they stand in for value rather than point toward it.

I say this without pretending innocence. I built my career on it. I scaled an org from a handful of people to a department where the consequences of failure were external and immediate. Single-digit attrition for nearly a decade. Engineers mentored to senior, staff, and lead. I built the ladders, ran the calibrations, defended the headcount. I worked for people I respected inside organizations that gave me real problems and real freedom. Some of the hardest and best work of my career happened inside those structures.

But that is not the end of the story. What sharpened my view was stepping away and watching what became possible when the translation layers fell. I left and built four products from zero. Smallest teams possible. AI-driven workflows. No layers. No coordination overhead. Decisions made close to the work. Context retained rather than re-narrated. Waste harder to disguise. I shipped faster, at higher quality, and learned more in nine months than in any comparable stretch of my career. High-quality work moved with a speed that many larger organizations would describe as unrealistic right up until they saw it happen.

I do not mean that scale is unnecessary. It is necessary. Complexity is real. Regulation is real. Operational rigor is real. I have spent too much of my career in consequential systems to romanticize improvisation. But there is a difference between real complexity and inherited complexity. Between stewardship and drag. Between leadership and accumulation.

That contrast is the foundation of this book.

The compression is here.

The team that ships fine with six instead of eight. The manager role that vanishes through attrition and nobody notices. The macro scale pretends nothing has changed.

We built elastic infrastructure. Auto-scaling compute. Databases that flex with demand. Entire platforms on the principle that fixed capacity is waste. Then we staffed the organizations that build those platforms with fixed hierarchies, fixed headcounts, and a career ladder so rigid that the only measure of progress is a title that maps to a comp band that maps to a calibration that maps to a negotiation the engineer will never see.

We made the machines elastic and the humans brittle. The career ladder persists not because it develops people but because it is how the organization sees itself. The structure is self-preserving. It will be compressed from outside, by economics, by tools, by the startups that prove it was never necessary. Or it will not be reformed at all.

This book argues for the compression. Not downsizing. Resizing. An org that composes and decomposes around actual problems and more importantly around massive latent value unlocks in the form of new products, features, and internal efficiencies. That makes work legible through systems, not through layers of interpretation. Deployment frequency. SLO adherence. Error budgets. User adoption. The work speaks for itself when you build the systems to listen.

That replaces the career ladder with visible, system-measured impact.

That replaces the team hierarchy with fire teams of two pairs. Expert Scalers and Slop Cannons. Deep-system thinkers and high-velocity builders. Operating within streams of subject-matter expertise: security, networking, user experience, payment rails, machine learning. Not teams with managers. Currents with practitioners.

Think of each pair as a container. Two processes: one Slop Cannon, one Expert Scaler. The container is elastic. Deployed to finance one week, sales the next. Composed with other containers for a large problem. Distributed for discovery across verticals. Handed an MVP and given a week. They flow to the highest-value problem. Loosely coupled. High impact. Stateless in their organizational attachment. Deeply stateful in their expertise.

Not anarchy. Orchestration.

Alignment follows structure. When a fire team describes its own output against defined facets -- did this expand the quality or quantity of external products? Internal tools? Did it generate compounding business intelligence that did not exist before? -- you create an exhaust that captures effort in terms the business can evaluate. That exhaust replaces the middle manager’s subjective interpretation of who contributed what. The evaluation becomes deterministic. The subjectivity drops out.

Build incentives on that exhaust and you create a team of value-seeking missiles. Every person, at every level, aligned to the same question: did this move the needle? Malinvestment -- the hierarchical instinct to expand scope and headcount without generating material impact on equity valuation or revenue -- becomes visible and accountable. The old career ladder rewarded accumulation. The compressed org rewards creation.

A compressed system still needs leadership, but a different kind. Not endless translators. Not custodians of abstraction. High-context, low-drama leaders who hold context across moving efforts, route talent toward consequential problems, resolve conflicts quickly, and maintain coherence without rebuilding bureaucracy as a coping mechanism. The person who understands the work talks to the person who funds the work. Everything between them is overhead.

The days of ceremonial, esoteric knowledge are ending. The cost of knowledge is approaching zero. What cannot be automated is the ability to look at a system and know what doesn’t need to be built at all. Systems design over vertical trivia. Latticed skill over narrow expertise. The person you want on your fire team, you will know them not by how quickly they implement code but by how quickly they identify the code that doesn’t need to be written.

This is the age of the generalist who goes deep on purpose. The builder who understands the landscape at an intuitive level because they have operated in enough of it to see the shape.

When I first got into software engineering, there was a term for the rare person who composed both sides of this coin. You would come to them with an idea on a Friday, and by Monday they would have an end-to-end working prototype. They were called 10x engineers. Unicorns. The industry treated them as anomalies. But what they were was a preview. They earned their place at the top of every hierarchy by delivering exactly what the future organization must: creative, high-value output at a velocity the old structure could not process. The Expert Scaler and the Slop Cannon are the 10x engineer split into two complementary halves. The fire team is the structure that composes that output without waiting for a unicorn to walk through the door.

The opportunity is larger than efficiency. Efficiency is the least interesting part of it.

The real opportunity is that the compressed organization has room for people the old one didn’t know how to value. The junior engineer with taste and drive who was told to wait three years before having an opinion. The subject-matter expert from sales, from finance, from operations, who understands the problem better than any engineer in the building but was never given a seat because they couldn’t write code. The non-technical thinker whose instinct for what users need has been filtered through three layers of product management, diluted beyond recognition. In the compressed org, these people are not waiting for permission. They are on the fire team. Their knowledge is the raw material. The tools meet them where they are.

And for the engineer two years into their career who picked up this book wondering whether they matter yet: you do. More than you know. The old hierarchy told you to apprentice quietly and climb slowly. The compressed org does not care how long you have been in the industry. It cares whether you can see a problem clearly, build something that solves it, and know when to stop. If you can do that, you are not a junior engineer. You are a Slop Cannon.

Leadership in this world carries a different weight. Not the weight of forced meetings and calibration theater. The weight of mentoring. Of being the person whose taste is contagious, whose architectural instinct saves the team a month of wrong turns, whose ability to unlock a technology for someone who didn’t know it existed changes the trajectory of a product. The best leaders in the compressed org are not managers. They are expert guides and taste disseminators and technology unlockers. They make the people around them measurably better at seeing what matters and ignoring what doesn’t. That is the job. Everything else the old hierarchy called leadership was overhead.

That is uncomfortable if your identity is tied to administrative territory. It is liberating if your identity is tied to building organizations that actually move.

This book is for every person who has sat in a meeting like the one that opened this preface and thought: there has to be something better than this.

There is. It is smaller, faster, more honest. It is being built right now by people who got tired of waiting.

Every chapter is two acts. The first is a story. The second is a treatise. The characters are fiction. The pressures are not.

Let’s move.